1. This forum section is a read-only archive which contains old newsgroup posts. If you wish to post a query, please do so in one of our main forum sections (here). This way you will get a faster, better response from the members on Motherboard Point.

PLCC socket problems

Discussion in 'Embedded' started by Guest, Apr 16, 2011.

  1. Guest

    Guest Guest

    Hi

    Has anyone experienced problems with PLCC sockets where the IC contact
    seems intermittent and possibly affected by variations in temperature.
    We have a problem with a Z80 micro and PLCC socket where the micro
    seems to behave erratically and a socket problem is suspected.
    Changing socket manufacturers didn't solve the problem.
     
    Guest, Apr 16, 2011
    #1
    1. Advertisements

  2. I have had to 're-form' both socket contacts (cantilever, and
    bifurcated haispring) as well as flip chips over and carefully re-form
    their legs with a scribe.

    More problems with the chips I think. Particularly if they are handled
    roughly/frequently in partial tubes. They band into each other and the
    pins suffer. Mostly the corner pins.

    Try upgrading to bifurcated contacts and inspect the chips.

    You don't think a thermal spec is being violated do you?

    good luck
     
    1 Lucky Texan, Apr 16, 2011
    #2
    1. Advertisements

  3. Guest

    D Yuniskis Guest

    What led you to believe/suspect this? Do mechanical tweaks to
    the assembly *while* it is misbehaving cause noticeable changes?
    I.e., power down, reseat the device and power back up again, etc.
    And you're *sure* it's not software, right? :>

    Does "erratically" mean "totally crazy" or just "not as we expect
    it to behave"? I.e., if you were to write a piece of code that
    just ran a simple scope loop, would you expect it to *also*
    "behave erratically"? (i.e., this would be something easy to prove!)
    Are you sure the contacts on the Z80 aren't oxidized or
    otherwise contaminated with something that is preventing
    good contact?

    With a high quality ohmmeter, have you measured the
    resistance between the pin *at* the PCB foil and the
    device itself? This requires some careful positioning
    of "probes". You can also "gut" a package (one part is a
    small price to pay for an answer) and probe from the bond
    wires "out".

    Have you tried *soldering* the device to the board (make or
    buy a little adapter that lets you solder the PLCC to the
    adapter and the adapter to the circuit board)? This allows
    you to remove the socket from the equation entirely (at least
    while you attempt to identify the real problem)

    HTH
     
    D Yuniskis, Apr 16, 2011
    #3
  4. Guest

    larwe Guest

    You make this sound so easy! Decapsulating plastic parts to the
    leadframe - even without trying to preserve the silicon - is pretty
    darn difficult.

    I'd pick an EPROM or ceramic part in a compatible LCC footprint and
    crack the lid off - much easier.
     
    larwe, Apr 16, 2011
    #4
  5. Guest

    David Brown Guest

    For sockets that were used a lot, I've sometimes had to use a fine bent
    pin to pull out all the connection springs to stretch them a little -
    that has often helped.

    But most of the problems I've seen with PLCC sockets where when
    customers changed the chips themselves. I've seen what happens when
    people can't work the chip extractors, and use a large pair of pliers
    instead. And it doesn't matter how well you label the board and the
    chips, and tell people they only fit one way - with the help of a
    hammer, customers can always get the chip to fit the way they want it to.
     
    David Brown, Apr 16, 2011
    #5
  6. Guest

    Rocky Guest

    We once found the a similar problem that was caused because the
    sockets were gold plated and the ICs were tin/lead. The lead and gold
    form a non-conductive amalgam. It turned out that it was possible to
    buy tin plated sockets which solved the problem.
     
    Rocky, Apr 16, 2011
    #6
  7. Guest

    D Yuniskis Guest

    That doesn't tell you anything about the Z80 that is believed to
    be (at least part of) the problem!
     
    D Yuniskis, Apr 16, 2011
    #7
  8. Guest

    larwe Guest

    Your posting that I have requoted above appears to be asking the OP to
    do a continuity check from the leadframe to the PCB foil, nothing to
    do with the die.
     
    larwe, Apr 17, 2011
    #8
  9. Guest

    D Yuniskis Guest

    Correct. His concern is that there *appears*/suggests that there
    is a problem in which "the IC contact seems intermittent". This
    involves two components -- an IC and a socket. *That* IC and
    *that* socket.

    Note that I never made any claims about the *die*. Rather, I was
    commenting on the PHYSICAL PART that the OP was having problems with.
    In particular, how it mates with the other physical component that
    the OP was having problems with -- the socket.

    Are you advocating he check continuity from ANY similarly sized
    PLCC package from ANY (possibly different) manufacturer and, from
    that, come to a conclusion about *this* PLCC package from *this*
    manufacturer??

    Why not test some *other* PLCC part in some other *vendor's*
    PLCC socket and come to conclusions that you hope apply to *this*
    part in *this* socket??

    If you think you can come to conclusions using a *different*
    PLCC component than the one in question, then just buy a PLCC
    test harness and plug *that* into the socket in question.

    Google (images) for "plcc plug". You'll find lots of off the
    shelf solutions for this problem (I have a PLCC68 in my cable bin).
    Buy one. Probably $100 (less than an hour of your time).

    And, from that, you will know whether *that* PLCC harness/plug works
    in *that* socket -- and very little more about why the Z80 PLCC
    appears *not* to.
     
    D Yuniskis, Apr 17, 2011
    #9
  10. Guest

    D Yuniskis Guest

    This is usually a longer term problem caused by the corrosion
    ("galvanic corrosion") resulting from the dissimilar metals
    being in electrical contact with each other.
     
    D Yuniskis, Apr 17, 2011
    #10
  11. Guest

    larwe Guest

    Poor debugging logic. If he tests a shape-compatible LCC part and it
    does have a problem, then the case is more or less proven. You are
    asking him to jump straight to a process that ideally requires fuming
    nitric acid and a fair amount of skill. Note: Not that I necessarily
    believe the problem has been correctly diagnosed. A quick squirt of
    freeze spray and/or a puff from a heat gun would probably help him
    learn a lot. At the speed of a Z80, I wouldn't expect serious issues
    with an in-spec socket and in-spec chip body.
     
    larwe, Apr 17, 2011
    #11
  12. As has been stated before in this thread...your idea to change the part
    in evaluation is nonsense. It really does need to be his part in his
    socket.

    To bare the lead frame of a PLCC chip does _not_ require working
    with hazardous chemicals. A piece of fine grit carborundum abrasive
    cloth or paper laid on a nice flat surface can be used to sand off the
    top of the PLCC chip in short order.
     
    Michael Karas, Apr 17, 2011
    #12
  13. Guest

    malcolm Guest

    We may have seen that - google fretting corrosion. IIRC.

    There are some vendors of pre lubricated PLCC sockets, but not easy
    to source.
    eg
    http://media.digikey.com/pdf/Data Sheets/Mill Max PDFs/Standard PLCC Sockets pg.93, Series 940.pdf

    We changed process to include a contact lubricant on the PLCC
    devices, and have since had no reports.
     
    malcolm, Apr 17, 2011
    #13
  14. Guest

    D Yuniskis Guest

    He's already tried a different manufacturer's socket. So, either
    two manufacturers have produced defective parts *or* the problem
    lies with his Z80's (or, the problem is "otherwise").
    No, you only need fuming nitric acid if you want to decap the
    device with the intent of salvaging the die. I had stated that
    the component could be *sacrificed*. His assertion is that the
    problem is not something internal to the package -- so, assume the
    die, bond wires, etc. are working perfectly... he assumes the
    problem is with the connection between "chip" and socket.

    He has stated that he has already tried another socket vendor.
    And the results are unchanged. So, if the problem *does* lie
    with the chip+socket, then it has to be THE *chip* that is tested,
    not some other that could be manufactured differently, etc.

    (e.g., what if the contacts are oxidized, enameled, etc. Or, the
    dimensions of the leads and/or chip body are not "to spec")

    Testing some other device's "connectivity" with the socket tells
    him nothing about the device that he is actually *using*.
     
    D Yuniskis, Apr 17, 2011
    #14
  15. Guest

    D Yuniskis Guest

    Exactly. An end mill -- or a surface grinder if you are *really*
    skillful -- will make quick work of shaving the top off the
    device. The trick will be ensuring that you don't significantly
    impact the position and retention of the lead frame within the
    *remaining* bits of epoxy. As such, you might have to mill off
    different parts of different "samples" so that what remains is
    still "testable" (*not* the electronics but just the lead frame).

    Recall the Z80 is a rather small die -- compared to the size of
    the PLCC. So, you have a fair bit of "mass" of lead frame that
    can remain embedded in non-abraded epoxy while still exposing
    "enough" for a test probe.

    My money is still on something *other* than a bad *mechanical*
    fit -- which is what the "connection" hypothesis seems predicated
    upon (unless the sockets are deforming in the board preheat
    cycle -- is this SMT or thru-hole assembly?)
     
    D Yuniskis, Apr 17, 2011
    #15
  16. Guest

    Glenn Guest

    Do not use sockets. Socket usually sucks over about 10MHz ;-)

    Instead use in-circuit upgrade?

    PS: Could it be a faulty PCB/via hole?
     
    Glenn, Apr 17, 2011
    #16
  17. Guest

    D Yuniskis Guest

    Z80 has nothing internally to upgrade. I'd guess the socket
    is there to:
    - allow SMT component (the Z80) to be used on a thru-hole PCB
    - allow ICE connection for development
    - allow easy access to all the I/O's for cheap ICT, etc.
    - allow optional daughterboard connection "on the cheap"
    etc.
    OP hasn't given enough information about his implementation,
    the actual symptoms observed, or the actual conditions under
    which it works/fails. The focus on "bad socket" (despite the
    fact that two different vendors' parts exhibit the same symptom?)
    seems either premature or there is a lot we haven't been told...

    (e.g., did the design *ever* work? Is the oscillator running
    properly? etc.)
     
    D Yuniskis, Apr 17, 2011
    #17
  18. Guest

    Dennis Guest



    We had this problems with SMD sockets, different socket manufacturers didn't
    fix the problem. Changed to a through hole socket fixed the problem. an
    extreme fix but it seemed to work for us.
     
    Dennis, Apr 17, 2011
    #18
  19. Guest

    D Yuniskis Guest

    Hi Dennis,

    How did you determine that the sockets were at fault?
    As you said, a bit "extreme" to alter the layout (so,
    presumably, you had convinced yourselves that this
    *was* the problem before undertaking such a big step).

    Were you ever able to determine what *aspect* of the
    sockets was the problem? E.g., deformation during
    assembly, bad wash, "made out of spec", too flimsy,
    contaminants, etc.? Or, perhaps some subtlety of
    the *layout* that changed when you moved to the
    thru-hole layout? Did you ever get any feedback from
    socket vendor(s) regarding the problem?

    As with the OP, did the fault appear to be temperature
    correlated?

    I.e., do you now have a "policy" of *never* using SMD
    sockets in designs? Or, have you been able to qualify
    *when* they can be used and when they shouldn't?
    (device size, types of signals involved, etc.)

    Out of curiosity, what size socket and what sort of
    device (technology) was *in* the socket?
     
    D Yuniskis, Apr 17, 2011
    #19
  20. Guest

    Dennis Guest


    The problem could be "fixed" by tweaking the contacts or squeezing the sides
    of the socket in by hand. ie Squeeze it works, let go it fails.......

    Poor cleaning was something we suspected but ruled out. We also looked at
    oxidation of the contacts on the package. The socket frames deformed
    slightly when the ic was inserted.
    Our volume was so small we didn't worry about conversing with the socket
    manuf.


    Probably not.
    I prefer to not use sockets anywhere if possible.


    IIRC 128 way Lattice pld or similar, it was about 10 years ago.
     
    Dennis, Apr 18, 2011
    #20
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.